Company Projects

AI Software Developer - DeukSpine Institute

• Figure out the ways in which AI can be used to solve the issues faced by patients and improvise user experience.

• Research and develop strategies to enhance business utilizing Generative AI, NLP and Data Science.

• Projects:
  • Healthcare Chat Bot - Deuk Bot
    • Currently training and creating the knowledge base for the chatbot using Google Dialog flow and Google cloud API’s.
    • Exploring migration factors and integration capabilites of Dialog Flow with various chat giants like ChatGPT , PaLM etc.
    • Engineered a data pipeline to streamline comments from the YouTube channel and a assist in the development of Knowledge base for the chatbot.

Embedded Software Developer - Sigsenz Technologies Pvt Ltd

• Was intially hired as an intern to work on the overall aspects of Edge IOT and AI.

• Considered my work, was converted to Full-time to work on Fleet and Driver analytics Solution

• Projects:
  • Driver Registration Application
    • Created a python based application for Driver Face Registration
    • Worked on face recognition using FACENET + UltraLight face detector
    • Utilised Azure Cognitive API’s for 2 level authenticaation of Face
    • Added features like face absense detection, OTA upddate etc to the software
  • Office and Industrial Surveillance system
    • Designed a simple software for food industry based on face recognition with a telegram alert system
    • Added multi camera support with camera based region detection
  • Driver Data Analytics
    • Started with data annotation for the various real time images using Azure cognitive services
    • Trained multi-object detection which included Facial mask, mobile-usage, seatbelt and drowsiness detection.

Data Scientist - AindraLabs Pvt Ltd

• Projects:
  • Reverse Vending Machine
    • Utilsed background subtraction methodology to detect the bottles on the moving conveyer belt.
    • Considered using Centroid tracker for the bottle tracking for robust results
    • Trained an Intra-class classifier consisting of 150 classes
    • Designed a 360 degree barcode scanning system
  • Wood Log detection and Recognition system
    • Created a robust highly augmented dataset for a outdoor environment by applying various luminosity effects using gamma reduction to match the outdoor luminosity effects
    • Trained a attention based convolutional model to further enhance the recognition capabilities of a simple CNN classifier
    • Looked into the visualization of CNN learning using GRAD-CAM
  • People Analytics
    • Optimised a deep learning people detector model using OpenVino and hence achieved a overall FPS of 25
    • Deployment of computer vision application on the remote client side
    • Trained a staff vs customer classifier using person reidentification network OSNet

---